论文部分内容阅读
【目的】地位指数法是森林立地质量评价常用的一种方法。采用广义代数差分法建立适用于杉木人工林的动态地位指数模型。【方法】利用福建省将乐县国有林场杉木人工林的24个固定样地连续观测数据和20株杉木优势木树干解析数据,基于Bertalanffy-Richards模型、Lundqvist-Kolf模型和Hossfeld模型3个经典的生长方程,以广义代数差分法对杉木人工林构建了6个动态地位指数模型。模型比较时综合考虑了统计学和生物学特征,通过统计分析和图形分析筛选出最佳的模型。【结果】构建的6个动态地位指数模型都具有良好的拟合优度,调整后的决定系数都在0.9左右。基于Hossfeld生长方程,选择a=b_1+X和b=b_2/X作为与立地有关的参数推导的模型确定为最佳模型,推荐采用该模型对将乐县国有林场人工杉木林进行优势树高生长预测和立地质量分类。【结论】广义代数差分法建立的动态地位指数模型具有较好预测性能,说明广义代数差分法在推导地位指数模型时是准确而有效的。在选择最优生长模型时不仅要考虑统计分析,还应该进行图形分析,从而选出满足统计学以及生物学特征的模型。
【Objective】 The index of status is a commonly used method for forest site quality evaluation. The generalized algebraic difference method was used to establish the dynamic position index model suitable for Chinese fir plantation. 【Method】 Based on the data from 24 continuous plots and 20 analytical trees of dominant tree species in the Chinese fir plantation in Jiangle County, Fujian Province, based on Bertalanffy-Richards model, Lundqvist-Kolf model and Hossfeld model, Growth equation, the generalized algebraic difference method for the Chinese fir plantation constructed six dynamic status index model. The model was comprehensively considered the statistical and biological characteristics, through the statistical analysis and graphical analysis to filter out the best model. 【Result】 All the 6 dynamic status index models constructed had a good goodness of fit and the adjusted coefficients were all around 0.9. Based on the Hossfeld growth equation, the best model was selected as a model derived from site-dependent parameters by selecting a = b_1 + X and b = b_2 / X. It is recommended that this model be applied to the height growth of dominant tree in Artificial Cunninghamia lanceolata Forestation in State- Prediction and site quality classification. 【CONCLUSION】 The dynamic index model established by generalized algebraic difference method has good predictive performance, which shows that generalized algebraic difference method is accurate and effective in deriving the index of position index. When selecting the optimal growth model, not only statistical analysis but also graphic analysis should be conducted to select a model that meets the statistical and biological characteristics.